Conclusion

After a long week of testing, we finally have all the data to make a qualified statement on PCI-Express scaling on Skylake, Windows 10 and AMD Radeon Fury X.

Just like in previous articles of this kind, there is no significant performance difference between x16 3.0 and x8 3.0 (and x16 2.0, which offers the same bandwidth). The average difference is only 1%, which you'd never notice. Even a bandwidth-restricted scenario, such as x16 1.1 or x8 2.0 offered by seriously old motherboards, only saw a small difference of around 4%. The same goes for x4 3.0, which is the bandwidth offered by the x4 slots on some recent motherboards. On recent platforms, the third x4 slot is usually wired through the chipset, which incurs an additional performance penalty that is definitely measurable and amounts to an additional 1%, but the difference is surprisingly small, much smaller than claims on the Internet talking about "crippled performance".

Real performance losses only become apparent in x8 1.1 and x4 2.0, where the performance drop becomes noticeable with around 6-10%. We also tested x4 1.1, though of more academic interest, and saw performance drop by up to 20%, an indicator that PCIe bandwidth can't be constrained indefinitely without a serious loss in performance.

Contrary to intuition, the driving factor for PCI-Express bus width and speed for most games is the framerate, not resolution, and our benchmarks conclusively show that the performance difference between PCIe configurations shrinks at higher resolutions. This is because the bus transfers a fairly constant amount of scene and texture data for each frame. The final rendered image never moves across the bus except in render engines that do post-processing on the CPU, which has gotten much more common since we last looked at PCIe scaling. Yet the reduction in FPS due to a higher resolution is still bigger than the increase in pixel data even then.

When looking at individual game results, the effects of constrained PCIe bandwidth vary wildly. Some games, like Metal Gear Solid V and Crysis 3, are virtually immune to limited PCI-Express bandwidth, and at the other end of the spectrum, you will find titles like Battlefield 4 that suffer a near 40% performance loss at worst.

In our previous article, we speculated that the advent of next-gen consoles would lead to games making more heavy use of bus bandwidth, but I'm happy to report that this doesn't seem to be the case, and recent titles behave similarly to old games when running on constrained PCI-Express slot configurations.

In the coming weeks, we will test GTX 970 SLI using various PCIe settings to investigate what happens in a multi-GPU setup.